bert4torch_config.json

{
  "attention_probs_dropout_prob": 0.0,
  "directionality": "bidi",
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 312,
  "embedding_size": 128,
  "initializer_range": 0.02,
  "intermediate_size": 1248,
  "max_position_embeddings": 512,
  "num_attention_heads": 12,
  "num_hidden_layers": 4,
  "type_vocab_size": 2,
  "vocab_size": 13685,
  "with_pool": "linear",
  "pool_strategy": "pooler"
}